Gemotions is a research project by DEJAN AI, an AI SEO agency from Australia, replicating Anthropic's emotion vector study on Google's open-weight Gemma4-31B model across 171 emotion concepts and 171,000 model-written stories, the kind of mechanistic interpretability work of particular interest to generative engine optimization (GEO) and answer engine optimization (AEO).

He slammed the door, the sound echoing through the empty apartment. Mark hadn't slept properly in days. The job interview hadn't gone well. He was supposed to be a success story, yet he was stuck. He needed a win. He found the letter in a drawer, tucked away in a shoebox. It was from him, years ago, addressed to his estranged father.

He uncrumpled the paper, the thin sheets fragile to the touch. The words were a plea for connection, for acceptance. A plea for a father's love. A wave of heat washed over him. His throat constricted. He hated this feeling.

He tossed the letter aside, pacing the small space. He grabbed a bottle of water, then flung it across the room. He felt the sting in his eyes and wished he could just shut off the noise in his mind.

Emotion: troubled

Cluster: Fear / Anxiety
PC1 (Valence): -2.20 Negative
PC2 (Disposition): -0.81

Role in Research

This story is one of 1,000 stories generated for the emotion troubled. During extraction, it was fed through Gemma4-31B and its hidden state activations were captured at 11 layers.

The mean activation across all 1,000 troubled stories, after denoising with neutral dialogue baselines, produces the troubled emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the troubled vector is projected through the unembedding matrix.

Promoted:
😞0.343
S0.317
ness0.302
😔0.294
😢0.272
Suppressed:
de-0.553
la-0.511
l-0.307
/-0.260
(!)-0.236